Hitachi

JP1 Version 12 JP1/IT Desktop Management 2 Distribution Function Administration Guide


3.1.6 Minimizing the load on the network

JP1/IT Desktop Management 2 transfers distributed software and various forms of management information over the network. To ensure that the system operates at its full potential, you need to take measures to minimize network congestion, for example by executing jobs when the load on the network is lightest. JP1/IT Desktop Management 2 provides a number of functions that help minimize the load on the network. This includes adjusting the network traffic during setup, and distributing software in segments. This subsection describes how to make the best use of network resources and distribute software as efficiently as possible. Efficient software distribution in JP1/IT Desktop Management 2 requires you to consider the settings and operations on managed computers, not just the managing server. For details about the tasks performed on managed computers, see Chapter 12. Performing Operations on the Agent.

Organization of this subsection

(1) Using relay systems

To transfer packages efficiently during remote installation, you need to understand the role of relay systems.

A relay system distributes the load on the network by reducing the number of managed computers that connect directly to the managing server. It also helps transfer packages to the managed computers under the relay system more efficiently by creating and saving copies of packages. Also, by combining a Transfer package to relay computer job with an Install package job, you can transfer a package in two stages.

The following figure shows the operation of a relay system during remote installation:

Figure 3‒3: Operation of a relay system during remote installation

[Figure]

A package specified in a Transfer package to relay computer job is stored on the relay system. The managing server will not redistribute the same package as long as it remains on the relay system. You can specify the storage period during packaging.

(2) Adjusting the number of hosts that can be connected concurrently

If the number of leased lines is limited or the network is unable to handle additional traffic, you need to adjust the number of hosts that can connect to the managing server at one time. There are two ways to adjust the number of concurrent connections on the managing server:

(a) Adjust the maximum concurrent connections during setup or in agent configurations

You can control network traffic in the setup of the managing server. From the menu, select Setup to display the Setup for Distribution by Using Remote Install Manager window, and then display the Server Customization Options page. For a relay system, use the Relay system settings page of the agent configurations.

Figure 3‒4: Server Customization Options page

[Figure]

The setting that governs the number of concurrent host connections is Number of JP1/IT Desktop Management 2 - Agent instances that can execute jobs concurrently. For a relay system, the relevant setting is Specify the number of job download requests to be executed concurrently, in the Processing settings for the relay system area of the Relay system settings page.

If a managing server executes jobs that target a large number of managed computers at the same time, the load on the network will increase dramatically. By specifying Number of JP1/IT Desktop Management 2 - Agent instances that can execute jobs concurrently, you can limit the number of lower hosts on which jobs can be processed at one time. For example, if you specify 20 for this setting, a job that remotely installs software on 300 computers is executed in 15 batches of 20 computers each.

(b) Adjust the maximum concurrent connections by user operations

During setup, you can specify a value that takes into account PC performance, the network configuration, and the operating environment. This provides a sufficiently reliable system in most cases. However, when large amounts of data are being transmitted, the network might become overloaded even if you control the amount of data being distributed. Ultimately, it might still be necessary for users to make adjustments.

■ Adjusting the number of computers targeted by a job

Suppose that you transfer a 5 MB package over a 64 Kbps line. If the line is 60% efficient, the transfer will take approximately 18 minutes. The transmission of a package from the managing server to the relay systems and from the relay systems to the managed computers takes place in parallel. Determine the number of computers to which a single job can be distributed by considering the load on the connection lines, the transfer speed, and the efficiency of the line.

■ Adjusting the number of concurrent connections by relay systems

To prevent a situation in which a large number of relay systems connect to the managing server at the same time, we recommend that you use split job execution. Although this method places additional burden on the user, it is an effective way of reducing the load on the network.

First, decide which relay systems connect concurrently to the managing server by considering the network traffic and other factors. Then, classify the managed computers under each relay system as a group. When you execute the job, leave an interval between job execution times for each group. The interval should be equivalent to the time it takes to process the job. In this way, you can avoid a situation in which a large number of relay systems connect to the managing server at the same time.

(3) Scheduled distribution

When you create a remote installation job, you can specify the date and time at which the job is executed. This serves as the transmission time of the package. You can transfer packages efficiently by specifying a time when network traffic is relatively light, such as at night.

If you use different jobs to transfer packages from the managing server to relay systems and from the relay systems to managed computers, you can specify different execution times for these jobs.

You can reduce the load on the network by scheduling package transfer in a considered way, rather than reducing the volume of data to be transmitted.

Note that software distribution cannot take place if the destination computer is in power saving mode. JP1/IT Desktop Management 2 - Agent (agent) is not active in this mode. If certain destination computers might enter power saving mode, take care when scheduling jobs that target those computers.

(4) Split package distribution

The software to be installed on a computer can be distributed in multiple segments. This is called split distribution. Split distribution offers the following benefits:

The following figure shows software distribution using split package distribution:

Figure 3‒5: Software distribution using split package distribution

[Figure]

(5) Multicast distribution

When distributing software normally, the number of packets sent from higher systems increases in proportion to the number of managed computers. However, JP1/IT Desktop Management 2 also supports a distribution method in which the higher system distributes software to a large number of specified managed computers by transmitting the packets for a job only once. This method is called multicast distribution, and can help reduce the total number of packets that need to be sent.

By specifying multicast distribution when you distribute software, you can reduce the amount of packet data sent by the managing server. This shortens the time required for distribution and reduces the load on the network. Multicast distribution is useful in the following situations:

The figure below shows software distribution using multicast distribution. You can use multicast distribution only between computers and their connection-destination higher systems.

Figure 3‒6: Software distribution by multicast distribution

[Figure]

For details about the system configuration and settings required to implement multicast distribution, see 3.2 Settings for multicast distribution.

(6) Suspending and resuming jobs

You can temporarily suspend job execution on a host where remote installation or package transfer is in progress. For example, you might schedule a job to be executed outside business hours. In this case, if the job has not finished when the business day begins, you can suspend the job and resume it after business hours.

You can also distribute a specific job while distribution of other jobs is suspended. For example, if you need to distribute a virus definition file urgently, you can suspend other distribution jobs and prioritize distribution of the virus definition file.

(a) Mechanism for suspending and resuming jobs

You can suspend and resume jobs by:

  • Issuing an instruction from the managing server that directs relay systems or management relay servers to suspend or resume distribution

  • Issuing an instruction to the managing server directing it to suspend or resume distribution

  • Executing the dcmsusp command on the managing server

For details about the dcmsusp command, see 14.17 dcmsusp.exe (suspending and resuming file transfer).

You can suspend file transfer in units of 4,096 bytes.

■ Issuing an instruction from the managing server that directs relay systems or management relay servers to suspend or resume distribution

You can direct relay systems to suspend or resume distribution by executing a Suspend file transfer job or a Resume file transfer job. As the destination of these jobs, specify the higher system that is directly connected to the hosts on which you want job distribution to be suspended.

The following figure shows the operation of the system during execution of a Suspend file transfer job:

Figure 3‒7: System operation during Suspend file transfer job

[Figure]

  1. From the managing server, execute a Suspend file transfer job to the relay system or management relay server.

  2. The destination relay system or management relay server enters suspended status.

  3. File transfer is suspended between the relay system or management relay server in suspended status and the systems directly under it, and the status of jobs destined for these lower systems changes to Suspended.

The following figure shows the operation of the system during execution of a Resume file transfer job:

Figure 3‒8: System operation during Resume file transfer job

[Figure]

  1. From the managing server, execute a Resume file transfer job to the relay system or management relay server in suspended status.

  2. The destination relay system or management relay server is released from suspended status, and the status of jobs destined for the lower systems changes to Resumed.

  3. Suspended file transfers are resumed, and the job status changes back to Running.

■ Issuing an instruction to the managing server directing it to suspend or resume distribution

You can use Remote Install Manager to issue an instruction to the local system directing it to suspend or resume distribution. To do so, from the Execute menu, select Suspend/Resume File Transfer and then Suspend or Resume.

The following figure shows the operation of the system when file transfer is suspended or resumed locally:

Figure 3‒9: System operation when suspending or resuming file transfer locally

[Figure]

  1. From the Remote Install Manager menu, select Suspend.

  2. The local system enters suspended status.

  3. File transfer is suspended between the local system and the systems directly below it, and the status of jobs destined for these lower systems changes to Suspended.

  4. From the Remote Install Manager menu, select Resume.

  5. The local system is released from suspended status, and the status of jobs destined for the lower systems changes to Resumed.

  6. Suspended file transfers are resumed, and the job status changes back to Running.

(b) Distributing jobs to suspended destinations

You can ensure that a job is distributed even when the system immediately above the destination is in suspended status. You can do this by selecting Distribute on the Job Distribution Attributes page when creating the job in the Create Job dialog box.

The following figure shows the distribution of a job to a suspended destination:

Figure 3‒10: Job distribution to suspended destination

[Figure]

When you suspend job distribution for a host, file transfer to systems directly below that host is suspended as soon as the host enters suspended status. However, file transfer will not be suspended for jobs for which the Distribute option was selected.

(c) Scope of suspension and resumption of job distribution

The scope of job suspension and resumption is as follows:

  • The following three types of jobs can be suspended:

    • Install package jobs

    • Transfer package to relay computer jobs

    • Send package, allow client users to choose jobs

  • You must use the managing server to create Suspend file transfer and Resume file transfer jobs.

(7) Using Client Control

Client Control is a function that allows you to start and shut down PCs remotely via a network. By using this function, JP1/IT Desktop Management 2 can remotely turn on remote PCs and install software at night, on weekends, and on holidays.

To use the Client Control function, the remote PC must support AMT or Wake on LAN, and also support automatic shutdown. Note that agents for UNIX cannot be shut down automatically after a job is executed. In addition, you cannot specify to shut down a computer for the following jobs executed on an agent for UNIX or an agent for Mac: Get system information from computer (UNIX) and Get software information from computer (UNIX).

The following figure provides an overview of remote installation using Client Control:

Figure 3‒11: Remote installation using Client Control

[Figure]

For notes on remote installation using Client Control, see 3.3 Settings for using Client Control.

(8) Using offline installation

You can install software on a computer with JP1/IT Desktop Management 2 - Agent installed that is used as a standalone device (an offline machine) without using the network. This process is called offline installation.

Offline installation involves creating media on the managing server that contains the files required for installation, and then physically transporting the media to the offline machine.

For details about how to perform offline installation, see 7.3.4 Offline installation.